Insights

We were promised a robot maid by now. Where humanoid robots have actually got to, when most of us will get one, and why the one worth having may not look like us.

The personal agents can now open your files, place your calls and spend your money. Meta has put one on a keychain. Before hiring the butler, it is worth asking what it knows and who it answers to.

A ChatGPT co-author shipped a model that cannot write a sentence, and developers are queuing to use it. Four cents a million words, an answer in a tenth of a second. Meet Jev.

Your firm remembers almost everything and can recall almost nothing. AI is beginning to close the gap. Every company will eventually need a brain of its own.

Hong Kong has taken the crown Switzerland wore for a century. What the country's banks must build — and quietly dismantle — to earn the next decade.

What the Enhanced Games actually tell us about the future of medicine.

The AI age may be creating a world unusually well-suited to Switzerland.

The case for taking the AGI prophets seriously — and what to do in Romandy while we wait to find out.

For thirty years, software did exactly what you told it. That deal has been quietly dissolving — and in the past six months it has become indefensible to pretend otherwise.

When a CEO simultaneously announces majority-AI code production and admits no one has solved AI for his industry, the headline number deserves more scrutiny than it received.

Open source is a supply chain, a talent signal, and a liability vector simultaneously. The companies treating it as one of the three will keep being surprised by the other two.

FINMA, FADP, and the EU AI Act now overlap on every fintech that lets a model touch client data. The companies treating it as one problem will outrun the ones treating it as three.

The CTO–CPO tension over roadmap ownership is the most reliably under-managed structural conflict in growth-stage companies. The strongest companies stop trying to resolve it and learn to operate inside it.

The Swiss senior-engineer market is structurally tight, getting tighter, and largely insensitive to the playbooks designed for it. The companies that win are not the ones paying most. They are the ones moving fastest, briefing most honestly, and treating retention as a hiring strategy.

Most Swiss organisations describe what they do as platform engineering. What most are actually running is shared infrastructure tooling under a more flattering name.

Eighteen months into widespread enterprise adoption, the empirical picture is sharper — and stranger — than either the boosters or the sceptics suggested.

AI workloads have quietly relocated significant portions of Swiss enterprise data outside the country, almost always without a board-level decision. The companies that recover sovereignty now will be the ones who treated it as an architectural choice rather than a political slogan.

Hybrid is not the hard part. The hard part is preserving technical decision quality when context fragments across time zones, Slack threads, and half-empty meeting rooms.

Switzerland's startup ecosystem is restructuring around AI, defence-tech, and a new generation of corporate capital. The deep-tech-spin-off playbook of the past decade no longer describes what is happening.

Agentic systems have crossed from demo to deployment in the past six months. The evaluation criteria most enterprise frameworks were built for do not yet cover the relevant failure modes.

The board no longer asks whether engineers are using Copilot. The board asks how it is being governed. The shift in the question is the most important data point in the room.

Geneva and Zürich's talent pools are small, expensive, and competed for by every employer with a Swiss postcode and a Bay-Area salary. The conventional sourcing playbook does not survive contact with the market.